Localization of an outdoor mobile robot considering the uncertainty model of the road curb

As one of the dominant features in the road environment, the curbs provide useful information for navigation. However, when the uncertainty model does not defined correctly, the road curb does not contribute to the estimated robot pose. In order to solve this problem, we propose the integrated localization scheme using curb features on the basis of uncertainty model of road curb. The uncertainty model for the extracted curb is quantitatively determined from experiments. Extended Kalman Filter is exploited for the fusion of curbs and DGPS measurements.